[EdLUG] Linux hibernate

Geetam geetam0 at gmail.com
Sat Sep 17 15:50:00 UTC 2022


I only know the theory of hibernating, so please fill in a bit of
hibernating knowledge that I do not have. (I am so old-fashioned that I
shut down and boot up 😬 )

When hibernating, the state of internal memory is stored on disk. Linux
uses the swap partition on disk for this. So your swap partition needs to
be large enough, i.e. at least as large as your internal memory - in your
case 32Gb

BUT WHAT HAPPENS when you have so much open on your desktop that your
internal memory wasn't big enough and some of the less active processes
have been swapped to swap space? (I am sure most of us avoid that, but that
is what swap space is for, isn't it?)

Say you have 40GB in use (32Gb in internal memory, 8Gb swapped to swap
space). Is hibernating going to try to save 40GB into swap space? That
won't fit into the 32Gb swap partition... Or is it only writing the 32Gb in
internal memory to the swap partition clobbering the 8Gb that was there?
Then the full state of your desktop cannot be restored.

Is this maybe why Andrew is not able to hibernate?

I am sure someone in the Linux world has thought about this, but I think we
have forgotten because we try to have so much internal memory that swaps
space is not used and the issue hardly ever arises these days. Still, I am
curious 🤔

Regards
Geetam

-- 
I don't know
...
I don't know what it is
...
I don't know what it is that I don't know

Isn't it beautiful


On Fri, 16 Sept 2022 at 08:04, Colin Shorts <colin.shorts at gmail.com> wrote:

> Hi Andrew, I'd be inclined to add a smidgen more swap to be on the safe
> side.
>
> I take it `sudo pm-hibernate` doesn't work (correctly) either? Are the
> kernel parameters getting set at boot (I think dmesg should say which
> device will be used)? Is your swap partition on lvm, are you mounting it
> using the uuid?
>
> I'm in the middle of redecorating my home office or I'd give it a bash
> myself 😃
>
> Cheers,
> Colin
>
> On Fri, 16 Sept 2022, 01:16 Andrew Smith, <asmith9983 at gmail.com> wrote:
>
>> Hi Colin
>> I have 32Gb of swap, with *htop* reporting  none used in normal
>> operation.
>> Suspend works OK. Given the minimal energy consumed overnight on suspend,
>> I may just continue using that until I can investigate further  how
>> *suspend-then-hibernate* is meant to operate by reading the
>> sourcecode of *systemctl. *
>> Andrew
>>
>>
>> On Wed, 14 Sept 2022 at 20:07, Colin Shorts <colin.shorts at gmail.com>
>> wrote:
>>
>>> I'm no expert on hibernate (I use suspend myself), but iirc you'll need
>>> at least as much free swap as system ram +used swap. 64GB doesn't sound
>>> unreasonable as a starting point assuming you've got enough space and
>>> considering how much Chrome can chew up.
>>>
>>> -Colin
>>>
>>> On Wed, 14 Sept 2022, 18:03 Andrew Smith, <asmith9983 at gmail.com> wrote:
>>>
>>>> Guys
>>>> For quite a while, I've been putting my system into suspend mode, which
>>>> was successful , apart from an odd occassion.  I now want to try saving
>>>> more energy
>>>> I've tried powering down, but would like it to restart in the same
>>>> state rather than with a new login, as I had with suspend.
>>>> I've tried "systemctl suspend-then-hibernate" from root CLI, but  I get
>>>> essentially  a fresh boot.
>>>>
>>>> Any suggestions ?
>>>> I'm running kernel 5.4.0-125-generic from Ubuntu, in 32Gb RAM, and
>>>> typically have a around hundred tabs open on Chrome.
>>>> Andrew
>>>> --
>>>> EdLUG mailing list
>>>> EdLUG at mailman.lug.org.uk
>>>> https://lists.edlug.org.uk/mailman/listinfo/edlug
>>>>
>>> --
>>> EdLUG mailing list
>>> EdLUG at mailman.lug.org.uk
>>> https://lists.edlug.org.uk/mailman/listinfo/edlug
>>>
>> --
>> EdLUG mailing list
>> EdLUG at mailman.lug.org.uk
>> https://lists.edlug.org.uk/mailman/listinfo/edlug
>>
> --
> EdLUG mailing list
> EdLUG at mailman.lug.org.uk
> https://lists.edlug.org.uk/mailman/listinfo/edlug
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.edlug.org.uk/pipermail/edlug/attachments/20220917/8b1ba341/attachment.htm>


More information about the EdLUG mailing list